Characterizing temporal variations of functional connectivity in resting-state

Abstract

The temporal variation in functional connectivity (FC) may convey important information about the integration and coordination of human brain. Recently, sliding-window analysis is a dominant approach to characterize temporal dynamics of FC, but there is still lacking an effective method to select the window size adaptively to cater for FC dynamics with different degrees of non-stationarity. In this work, we introduce a data-driven variable window selection method for estimating the time-varying correlation coefficient and apply it to investigate temporal variability of FC in resting-state fMRI. The results demonstrate that between-network FC exhibits a significantly larger temporal variation than within-network FC.